Encounter History Length with acoustic telemetry data

[esnook]

My study is a 3-year passive acoustic telemetry study of sea-run brook trout, focused on understanding their movement patterns within the system as influenced by environmental variables. We tagged fish in 5 samplings throughout the study. The maximum life of an acoustic tag that we have seen was about 400 days. We have detected 64 fish total on our receivers.

Our plan is to use a multi-state model in MARK. We joined the receivers into four ‘nodes’ or locations. Those are the different states (0-4). We’re especially interested in times when fish transition to node 4, which is also the least and most quickly visited node. Since we often have a fish move quickly between two nodes (from several minutes to a few hours), we decided to make the occasions equal to stages of the tidal cycle. So in general we have four tidal stages, and thus four occasions, in day. It is pretty important that our occasions be less than a day because we have seen fish traverse all four nodes in a day. To condense our detections, we took the mode of the node value within the tidal stage time interval. In the end, the encounter histories for our 64 fish end up being 3907 occasions in length. MARK apparently only accepts 240 occasions.
An few examples of our EH’s: …0000000000003333032333330000330332000000…
…000000000000300333333040404333333333333…
… 030000000000000000000000000000000000022000011000..

There are probably ways that we could cut this number down. There are a lot of zeros in general because fish weren’t detected constantly. For many fish, there are a lot of zeros at the beginning and end of the encounter history because they were tagged in the middle of the study, for example, and because the tag life is limited to 400 days. However, if we reduced the EH length to the max tag life we would still have an encounter history of 1600 occasions (400days*4occasions/day). Also, there is high chance that reducing our EH to fewer occasions would cause us to miss movement into node 4 altogether since we are using the mode of the node value to fill the state.

Any thoughts on how we might get MARK to accept an increased number of occasions (maybe using the Linux version)? Or any ideas on how we could whittle down the EH without missing those important transitions between states?

Thanks!
Erin

[Fish_Boy]

Now that I understand your project better (see post in viewtopic.php?f=1&t=1273) I think you can get a handle on your data by using shorter periods of record, at least initially to assess the parameters of interest. For example, you can take all your records from Spring periods and assess the parameters, then do summer, and fall spawning. Of course there will be annual variation in precise timing of migrations etc. but with concurrent temperature data you could synchronize the movements based on temperature using degree days or some similar metric to parse out some shorter time interval e.g., 240 is the maximum occasions 240 occasions /4 occasions per day = 60 day windows. Additionally, you can use covariates to keep track of different years of data when assessing things this way. As noted mentioned in my post in the aforementioned (max occasions) post, you can parse out your data any way you choose based on time intervals to assess your movement data to get an initial understanding of how the parameters of interest change. This gives you a work around for the time being. In theory as long as the smaller time periods of data are associated with a biological period where fish are expected to be behaving in a similar fashion (e.g., migrating, spawning, feeding, rearing etc.) I find the movement data to be robust enough to be definitive.